0

I would like to test in a script [Linux: CentOS 7] the availability of a remote machine. While TCP packages can blocked by a firewall on the way, ICMP packages may be possible and are not blocked by the default firewall settings. Thus, ping is no help for me

I tried traceroute with a limited number of hops. But I get always an exit code of $?=0 even if the target has not been reached within a given number of hops, e.g.,

traceroute -T -p 1234 192.168.123.45 -w 1 -q 1 -m 5

So, I am looking for a way how to get on the CLI a 'proper error' as exit code if a machine cannot be reached via TCP (for a given set of parameters)?

Solution/Duplicate

Cyrus pointed to Test from shell script if remote TCP port is open which solves my problem with netcat

reading from the tcp pseudo-device may block, when the host does not answer at all on tcp

   </dev/tcp/192.168.123.45/1234

thus this would work only for testing ports on reachable hosts. Unfortunately, I found no way to reliable timeout reading from such a pseudo-device (timeout -s4 2 </dev... does not work)

With netcat probably available on most setups, I am going now for a check like

nc -z -w 2 192.168.123.45 1234 && echo "host foo reachable on port bar" || echo "host foo not reachable on port bar"

with a two second timeout

Community
  • 1
  • 1
THX
  • 553
  • 2
  • 8
  • 18
  • What do you mean by "Since ICMP packages may be possible but TCP being blocked by a firewall on the way" ? Please clarify. – rakib_ May 26 '16 at 14:04
  • 2
    Only with bash: [Test from shell script if remote TCP port is open](http://stackoverflow.com/a/35338529/3776858) – Cyrus May 26 '16 at 14:08
  • 1
    does the `traceroute` give specific output for success vs fail? – webb May 26 '16 at 16:25
  • 1
    @Cyrus many thanks! I did not know that bash with tcp pseudo-devices is possible and I even do not need another tool. – THX May 27 '16 at 07:29

0 Answers0